Todd-Brown, K. | bioRxiv Science

bioRxiv2023

Open Soil Spectral Library (OSSL): Building reproducible soil calibration models through open development and community engagement

Soil spectroscopy is a widely used method for estimating soil properties that are important to environmental and agricultural monitoring. However, a bottleneck to its more widespread adoption is the need for establishing large reference datasets for training machine learning (ML) models, which are called soil spectral libraries (SSLs). Similarly, the prediction capacity of new samples is also subject to the number and diversity of soil types and conditions represented in the SSLs. To help bridge this gap and enable hundreds of stakeholders to collect more affordable soil data by leveraging a centralized open resource, the Soil Spectroscopy for Global Good has created the Open Soil Spectral Library (OSSL). In this paper, we describe the procedures for collecting and harmonizing several SSLs that are incorporated into the OSSL, followed by exploratory analysis and predictive modeling. The results of 10-fold cross-validation with refitting show that, in general, mid-infrared (MIR)-based models are significantly more accurate than visible and near-infrared (VisNIR) or near-infrared (NIR) models. From independent model evaluation, we found that Cubist comes out as the best-performing ML algorithm for the calibration and delivery of reliable outputs (prediction uncertainty and representation flag). Although many soil properties are well predicted, total sulfur, extractable sodium, and electrical conductivity performed poorly in all spectral regions, with some other extractable nutrients and physical soil properties also performing poorly in one or two spectral regions (VisNIR or Neospectra NIR). Hence, the use of predictive models based solely on spectral variations has limitations. This study also presents and discusses several other open resources that were developed from the OSSL, aspects of opening data, current limitations, and future development. With this genuinely open science project, we hope that OSSL becomes the driver of the soil spectroscopy community to accelerate the pace of scientific discovery and innovation.

bioinformatics↗

bioRxiv2019

Non-rainfall moisture: a key driver of carbon flux from standing litter in arid, semiarid, and mesic grasslands

Models assume that rainfall is the major source of moisture driving decomposition. Non-rainfall moisture (NRM: high humidity, dew, and fog) can also induce standing litter decomposition, but there have been few standard measurements of NRM-mediated decompositions across sites, and no efforts to extrapolate the contribution of NRM to larger scales to assess whether this mechanism can improve model predictions. Here we show that NRM is an important, year-round source of moisture in grassland sites with contrasting moisture regimes using field measurements and modeling. We first characterized NRM frequency and measured NRM-mediated decomposition in sites on the extreme dry and wet end of grassland systems: at two sites in the Namib Desert, Namibia (hyperarid desert) and at one site in Iowa, USA (tallgrass prairie). NRM was frequent at all sites (85-99% of hours that litter was likely to be wet were attributed to NRM) and tended to occur in cool, high-humidity periods for several hours or more at a time. NRM also caused respiration of standing litter at all sites when litter became sufficiently wet (>5% for fine litter and >13% for coarse), and contributed to mass loss, even in the Namib West site that had almost no rain. When we modeled annual mass loss induced by NRM and rain, and extrapolated our characterization of NRM decomposition to a final site with intermediate rainfall (Sevilleta, New Mexico, semiarid grassland), we found that models driven by rainfall alone underestimated mass loss, while including NRM produced estimates within the range of observed mass loss. Together these findings suggest that NRM is an important missing component in quantitative and conceptual models of litter decomposition, but there is nuance involved in modeling NRM at larger scales. Specifically, temperature and physical features of the substrate emerge as factors that affect the common microbial response to litter wetting under NRM across grasslands sites, and require further study. Hourly humidity can provide an adequate proxy of NRM frequency, but site-specific calibration with litter wetness is needed to accurately attribute decomposition to periods when NRM wets litter. Greater recognition of NRM-driven decomposition and its interaction with other processes (e.g. photodegradation) is needed, especially since fog, dew, and humidity are likely to shift under future climates.\n\nManuscript highlightsO_LINon-rainfall moisture (NRM; humidity, fog, dew) induces decomposition in grasslands\nC_LIO_LINRM decomposition depends on substrate type, and occurs at colder times than rain\nC_LIO_LIIncluding NRM (instead of rain alone) improved predictions of litter decomposition\nC_LI

ecology↗